Small Group Research
http://sgr.sagepub.com
Does Team Building Work?
Cameron Klein, Deborah DiazGranados, Eduardo Salas, Huy Le, C. Shawn
Burke, Rebecca Lyons and Gerald F. Goodwin
Small Group Research 2009; 40; 181 originally published online Jan 6, 2009;
DOI: 10.1177/1046496408328821
The online version of this article can be found at:
http://sgr.sagepub.com/cgi/content/abstract/40/2/181
Published by:
http://www.sagepublications.com
Additional services and information for Small Group Research can be found at:
Email Alerts: http://sgr.sagepub.com/cgi/alerts
Subscriptions: http://sgr.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations http://sgr.sagepub.com/cgi/content/refs/40/2/181
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Does Team Building Work?
Cameron Klein
Deborah DiazGranados
Eduardo Salas
Huy Le
C. Shawn Burke
Rebecca Lyons
Small Group Research
Volume 40 Number 2
April 2009 181-222
© 2009 SAGE Publications
10.1177/1046496408328821
http://sgr.sagepub.com
hosted at
http://online.sagepub.com
University of Central Florida
Gerald F. Goodwin
Army Research Institute
This research reports the results of a comprehensive investigation into the
effectiveness of team building. The article serves to update and extend Salas,
Rozell, Mullen, and Driskell’s (1999) team-building meta-analysis by assessing a larger database and examining a broader set of outcomes. Our study
considers the impact of four specific team-building components (goal setting,
interpersonal relations, problem solving, and role clarification) on cognitive,
affective, process, and performance outcomes. Results (based on 60 correlations) suggest that team building has a positive moderate effect across all
team outcomes. In terms of specific outcomes, team building was most
strongly related to affective and process outcomes. Results are also presented
on the differential effectiveness of team building based upon the team size.
Keywords:
team building; team performance; team development
Teams of people working together for a common cause touch all our lives.
From everyday activities like air travel, fire fighting, and running the United
Way drive to amazing feats of human accomplishment like climbing Mt.
Everest and reaching for the stars, teams are at the center of how work gets
done in modern life.
Kozlowski & Ilgen, 2006, p. 78
This quote, from a recent review of work-team effectiveness, exemplifies
the central role that teams play in our lives. Although many labels have
been applied to team-based forms of organizing (i.e., crews, teams, groups,
and collectives), these entities are essential to the accomplishment of organizational goals. Indeed, there is ample support in the literature for the
181
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
182
Small Group Research
contention that team-based forms of organizing are beneficial both to
organizations and to individuals. For example, Applebaum and Batt (1994)
reviewed 12 large-scale surveys and 185 case studies of managerial practices and concluded that team-based work leads to improvements in organizational performance on measures of both efficiency and quality. They
argued that team-based systems benefit workers because of the higher likelihood of job enhancement, autonomy, and skill development associated
with these systems. However, the simple existence of a team-based organizing structure is not enough to ensure that positive outcomes will result.
Teams must be nurtured, supported, and developed.
The Motivation for Understanding the
Efficacy of Team Building
There are three motivations for understanding the efficacy of teambuilding interventions in organizations. First, team building is one of the
most commonly applied group development interventions in organizations
today. It is widely used and comes in many forms, including outdoor experiential activities and indoor group process discussions. However, no one is
quite sure how and why these interventions work, or if they even work at
all. Considering the vast sum of money directed toward the development of
teams in organizations, it is important that practitioners (and researchers)
gain a better understanding of the effectiveness and boundary conditions of
team building.
Second, as there are many options available to organizations in the
pursuit of improved teamwork, it is important to determine whether team
building is a worthy choice. Some of these interventions are organizational
Authors’ Note: This work was partially supported by funding from the U.S. Army Research
Institute for the Behavioral and Social Sciences (Contract W74V8H-04-C-0025) and grant
SES0527675 from the National Science Foundation, awarded to Glenn Harrison, Stephen M.
Fiore, Charlie Hughes, and Eduardo Salas. All opinions expressed in this article are those of
the authors and do not necessarily reflect the official opinion or position of the University of
Central Florida, the Department of the Army, the Department of Defense, or the National
Science Foundation. Portions of the article were presented at the 2nd Annual Conference of
the Interdisciplinary Network for Group Research (INGRoup) held at Michigan State
University, in Lansing, Michigan. Correspondence concerning this article should be addressed
to: Dr. Eduardo Salas, Institute for Simulation & Training, University of Central Florida, 3100
Technology Parkway, Suite 132, Orlando, FL 32826, USA. Tel: (407) 882-1325; fax: (407)
882-1550; e-mail: [email protected].
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
183
or structural in nature and do not specifically target team member interactions (e.g., job redesign, selection systems, and group incentive and
performance management programs). In contrast, team-development interventions—those whose quest is to directly impact the functioning and
effectiveness of work teams—provide the focus for the current research
integration. These interventions, when properly conducted, can have a
positive impact on organizations. Consider a research study conducted by
Macy and Izumi (1993), who analyzed 131 studies of organizational
change. They found that interventions with the largest effects upon financial measures of organizational performance were team-development
interventions. That is, of all organizational interventions, those that focus
on team development had the largest effect on measures of financial
performance.
Third, beyond financial performance, it is widely understood that team
developmental interventions are key mechanisms that may be used to facilitate team effectiveness (Noe, 2002). Therefore, it is important to understand how these interventions are most effective. At a general level,
team-development interventions may include some form of team-training
or team-building activities. Although both types of team-development interventions are designed to improve team functioning and effectiveness (especially when the science of individual and team training is utilized; Salas &
Cannon-Bowers, 1997), team training and team building differ in important
ways (Tannenbaum, Beard, & Salas, 1992). Team training is skill-focused
(i.e., it is focused on gaining specific competencies), typically includes a
practice component, and is done in context. It is generally formal and
systematic. Team building, on the other hand, does not target skill-based
competencies, is not systematic in nature, and is typically done in settings
that do not approximate the actual performance environment. For our purposes, we define team building as a class of formal and informal team-level
interventions that focus on improving social relations and clarifying roles,
as well as solving task and interpersonal problems that affect team functioning. Team building works by assisting individuals and groups to examine, diagnose, and act upon their behavior and interpersonal relationships
(Schein, 1969, 1999).
In light of the above, this article investigates the efficacy of team building.
We begin with a brief review of conceptual issues and methodological issues
in team building. Next, the hypotheses for this research are presented.
Finally, the results of several meta-analytic integrations are discussed. First,
however, is the rationale for the present research.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
184
Small Group Research
Understanding the Need for an Update
It is an unfortunate indictment of the literature and practice in this area
that we are still searching for answers to questions posed by Beer (1976)
and Salas, Rozell, Mullen, and Driskell (1999). Namely, does team building result in positive outcomes? Why? Under what conditions? Upon a
careful review of the extent literature, it is clear that these questions need to
be examined more closely to clarify our understanding of the effectiveness
and boundary conditions of team building. Thus, this article reports the
findings from a series of meta-analytic investigations of research on the
efficacy of team-building interventions to update and extend the current
state of knowledge in the team-building domain. In total, this research will
examine three moderator variables in addition to making an overall assessment of the efficacy of team-building interventions. The current replication
is meant to be systematic, rather than direct in nature (e.g., Aronson,
Ellsworth, Carlsmith, & Gonzales, 1990). That is, rather than replicating
the exact analyses from the study by Salas and colleagues, the current
research will replicate and extend the earlier study in an effort to resolve
ambiguities and provide new insights for organizational stakeholders and
academicians alike concerning the effectiveness of team building.
Although the empirical evidence on team-building interventions is limited, a critical investigation of the available literature is warranted for several reasons. First, since the 1990s there has been an increasing incursion
of team-building interventions in organizations. Some of the most recent
trends in team building have taken these interventions into the kitchen and
even the wilderness. Second, many practitioners feel that these interventions are useful. However, a more careful exploration of the utilities and
strengths of these interventions would benefit practitioners now and would
benefit the development of these interventions in the long run. Finally, there
are enough data available to form estimates regarding some of the variables
that may moderate the impact a team-building intervention may have on
team outcomes.
There is a general consensus for the idea that the science of team training can be applied to enhance team functioning across organizations (e.g.,
Kozlowski & Ilgen, 2006; Salas & Cannon-Bowers, 1997). At the same
time, reviews of team effectiveness have noted the inconsistent findings for
the effectiveness of team-building interventions. Despite these inconsistent
findings, researchers such as Kozlowski and Ilgen have acknowledged the
potential that these interventions may have on shaping team development
and on improving team effectiveness. Given the increasing frequency of
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
185
team-building interventions, if researchers do not catch up and start doing
this research, an important window might be missed to constructively shape
the important practice in this area. Thus, examining team building further—
a topic that has not received recent sufficient attention in the literature—is
critical. In the next section, a number of conceptual and methodological
issues in team building are discussed.
Team Building
Conceptual Issues in Team Building
Originally designed as a group process intervention (e.g., Schein, 1969,
1999) for improving interpersonal relations and social interactions, team
building has evolved to also include a concern for achieving results, meeting goals, and accomplishing tasks (Payne, 2001). In the late 1990s, Salas
and colleagues (1999) described team-building interventions as extremely
popular and common. According to Beer (1976), there are four basic
approaches to team building, including: (a) a goal-setting, problem-solving
model; (b) an interpersonal model; (c) a role model; and (d) the Managerial
Grid (Blake & Mouton, 1964) model. However, this initial conceptualization has since been reconsidered.
Refinements to this four-pronged system began with the Managerial Grid
model being dropped as a distinct team-building approach. In addition to
dropping the Managerial Grid model, modern conceptualizations have separated the goal-setting and problem-solving approaches, using Buller’s
(1986) problem-solving component as a distinct approach. As discussed by
Buller, team-building models rarely exist in pure form. That is, the interventions reported in the literature usually involve elements from several or all of
the models. As a result, he proposed a general problem-solving model that
follows Dyer’s (1977) problem-solving framework. This model incorporates
a focus on task or interpersonal issues, goal setting, and role clarification,
depending on the nature of the specific problems identified for the group
under investigation. Moreover, the problem-solving approach to team building is said to subsume each of Beer’s (1976) components, and, as perhaps
evident by its title, emphasizes the identification of major problems in the
team. All told, these modifications have added clarity to the investigation
and implementation of team building. Unfortunately, this newfound conceptual clarity may not have come soon enough for many investigators who had
previously sought to assess the efficacy of team building.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
186
Small Group Research
Perhaps because of the conceptual confusion that existed in this area,
many reviews of team building did not include the same articles. For example
two noteworthy reviews (DeMeuse & Liebowitz, 1981; Woodman &
Sherwood, 1980) had only 14 studies in common out of a total of 66 studies
reviewed. A subsequent review by Buller (1986) included only 9 studies, 3 in
common with Woodman and Sherwood (1980) and 6 in common with
DeMeuse and Liebowitz. One reason for the apparent lack of consistency in
the articles chosen for reviews is that team building has been an ill-defined
concept (Buller, 1986). At the time of these early reviews there was no
agreed-upon operational definition of the intervention. Taken together, there
is now a consensus position that there exist four distinct models of team
building. Although combinations of these approaches are common, the
models include goal-setting, developing interpersonal relations, clarifying
roles, and creating additional capacity for problem solving (Beer, 1976;
Buller, 1986; Dyer, 1987; Salas et al., 1999). Table 1 describes each of the
four models or components of team building in more detail.
Despite the recently established consensus on team-building components, there have been other problematic issues that have persisted for
researchers and practitioners of this topic. Specifically, many early efforts
were plagued by a number of methodological issues. A few of these are discussed in the next section.
Methodological Issues in Team Building
Early reviews of team building described both a lack of extensive research
on the issue and trepidation concerning the methodological rigor of published
studies (Buller, 1986; DeMeuse & Liebowitz, 1981; Tannenbaum et al.,
1992; Woodman & Sherwood, 1980). That is, even if one could get beyond
the disagreement concerning operational definitions of team building,
much of the previous team-building research is characterized by methodological flaws (i.e., study design and measurement issues). For example,
Buller (1982) found that more than half of the reported team-building studies
employed pre-experimental designs—designs that do not allow for causal
inferences. Also problematic, there was often no attempt to disentangle the
effects of team building from other interventions that may be in process
within an organization. Although this can be a common problem for field
research in general, it has been particularly problematic in the team-building
domain.
Tannenbaum and colleagues (1992) highlighted another methodological
flaw in the team-building research. Specifically, they pointed out that there
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
187
Table 1
Models/Components of Team Building
Component
Goal setting
Salas, Rozell, Mullen,
& Driskell, 1999
Salas, Priest, &
DeRouin, 2005
Emphasis: Setting objectives
and development of
individual and team goals.
Team members: Become involved
in action planning to identify
ways to achieve goals.
Designed to strengthen team member
motivation to achieve team goals
and objectives.
By identifying specific outcome
levels, teams can determine what
future resources are needed.
Individual characteristics (e.g., team
member motivation) can also be
altered by use of this intervention.
Interpersonal Emphasis: Increasing teamwork
Based on the assumption that teams
relations
skills (i.e., mutual supportiveness,
with fewer interpersonal conflicts
communication, and sharing
function more effectively than
of feelings).
teams with greater numbers of
Team members: Develop trust in
interpersonal conflicts.
one another and confidence
Requires the use of a facilitator to
in the team.
develop mutual trust and open
communication between team
members.
As team members achieve higher
levels of trust, cooperation, and
cohesiveness, team characteristics
can be changed as well.
Role
Emphasis: Increasing communication Defines the team as comprising a set
clarification
among team members regarding
of overlapping roles.
their respective roles within
These overlapping roles are
the team.
characterized as the behaviors that
Team members: Improve
are expected of each individual
understanding of their own and
team member.
others’ respective roles and duties Can be used to improve team and
within the team.
individual characteristics (i.e., by
reducing role ambiguity) and work
structure by negotiating, defining,
and adjusting team member roles.
Problem
Emphasis: Identifying
Buller’s (1986) problem-solving
solving
major task-related problems
component subsumes aspects from
within the team.
all of the components described by
Team members: Become involved
Beer (1976).
in action planning, implement
Team members practice setting goals,
solutions to identify problems
develop interpersonal relations,
clarify team roles, and work to
(continued)
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
188
Small Group Research
Table 1 (continued)
Component
Salas, Rozell, Mullen,
& Driskell, 1999
and to evaluate those
solutions.
Salas, Priest, &
DeRouin, 2005
improve organizational
characteristics through
problem-solving tasks.
Can have the added benefit of
enhancing critical-thinking skills.
has been a reliance on measuring the effectiveness of team-building interventions with process measures. Although implicitly appealing, improvements in processes can not always be linked to improvements in team
performance (e.g., Porras & Wilkens, 1980). For example, team performance is typically fashioned by additional environmental and/or organizational characteristics and contingencies that are out of the volitional control
of team members. Stated differently, team members are active participants
in the enactment of team processes, but must also interact within the larger
system to produce more distal performance outputs. Team processes may
also be impacted by the larger organizational environment, but not likely to
the same degree as team performance outputs.
As a final methodological concern with previous team-building
research, there has been an overreliance on subjective indicators of group
or organizational performance criteria as dependent measures (Tannenbaum
et al., 1992). Though this type of information may be interesting and relevant to measuring participant satisfaction and other affective outcomes that
may be impacted by team-building interventions, it is often not concrete
enough to allow for accurate predictions of the performance outcomes of
team building.
A Recent Advancement in Team-Building Research
Despite the conceptual and methodological issues associated with evaluations of team building, one recent effort has represented advancement
over previous reviews by empirically investigating the effectiveness of
these interventions. Specifically, Salas and colleagues (1999) responded to
a call by Buller (1986) to use the primary focus of the intervention
(i.e., goal setting, interpersonal relations, role clarification, and problem
solving) as a potential moderating variable. The results of their study failed
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
189
to indicate a relationship between the combined set of team-building interventions and team performance (r = .01, k = 16 effect sizes). Moreover, of
the four components of team building, only role clarification proved to be
effective, as judged by both objective (r = .71) and subjective (r = .75)
accounts of performance. Interestingly, for objective and subjective measures there was a nonsignificant effect of goal setting (r = −.06 and −.11,
respectively), interpersonal relations (r = −.38 and −.04, respectively), and
problem solving (r = −.31 and .09, respectively). Also examined in this
study were the potential moderating influences of the source of the criterion
measurement of performance (i.e., objective vs. subjective), team size, and
training duration. For objective measures of performance, there was no evidence of a relationship between team building and performance (r = −.04,
k = 8); for subjective measures, there was a small positive relationship
between team building and performance (r = .14, k = 8). Concerning team
size, the results of their study suggested that the effects of team building on
performance decreased as a function of the size of the team (r = −.34).
Finally, there was a slight tendency for the effects of team building to
decrease as a function of the duration of the intervention (r = −.20).
Taken together, this study enhanced our understanding of the efficacy of
team building. However, there were a number of limitations associated with
it—limitations that now necessitate the need for additional inquiry. For
example, the amount of data analyzed in this study was relatively modest.
Specifically, the research findings presented were based on only 16 effect
sizes, and thus, it is difficult to discern, with any degree of confidence, the
actual effectiveness of these interventions. Moreover, the findings derived
from the Salas and colleagues’ (1999) study did not necessarily reflect the
findings from existing narrative reviews. Finally, their study left many
questions unanswered regarding the potential moderating impact of other
relevant variables. As an organizing tool provided to summarize the literature in this area, Table 2 provides a summary of five previous investigations
into the efficacy of team building, and includes the number of articles
reviewed, the years spanned, and other noteworthy features.
In summary the current research was initiated to provide an updated
meta-analysis of the team-building literature. Phrased in the form of questions, an examination of these issues will help clarify our understanding of
the effectiveness of team building and lead to hypothesized relationships
involving the effectiveness of team building, including an investigation of
specific moderators. A model that serves to graphically illustrate these
hypotheses is presented in Figure 1.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
190
Small Group Research
Table 2
Summary of Previous Team-Building Reviews
Review
Study
Details
Woodman & 30 articles
Sherwood
(1980)
Qualitative
data
Summary of
Findings
TB elicits positive affective
reactions.
The linkage between TB and
work group performance
remains largely unsubstantiated.
Years spanned: TB is more commonly
1964–1978
conducted with management
teams than groups lower
in organizational hierarchies.
TB is more commonly conducted
with intact and established
work teams than new groups.
Affective reactions as dependent
measures are used more often
than objective performance data.
DeMeuse & 36 articles
TB is described as having great
Liebowitz
promise for improving employee
(1981)
Qualitative
attitudes, perceptions, behaviors,
data
and organizational effectiveness.
Eighty-seven percent of the 68
Years spanned:
evaluations indicated positive results.a
1962–1980 Due to the lack of rigorous research
designs, firm conclusions
concerning the effectiveness of these
interventions could not be made.
Buller
9 articles
TB must be more carefully defined.
(1986)
More rigorous experimental
Qualitative
designs should be used.
data
There are numerous methodological
flaws in previous TB studies.
Years spanned: Therefore a clear assessment of the
1964–1981
TB and task performance
relationship had yet to emerge.
Noteworthy
Features
Distinguished between team
development and T-group
or sensitivity training.
Made subjective assessments
of the internal validity
of the studies reviewed.
Coded studies according to
research design, sample size,
multiple dependent variables,
and duration of the TB
intervention.
Presented and described a
general problem solving
approachto TB that added to
Beer’s (1976)
four-component model.
Argued that Beer’s
classification is difficult to
use in practice because TB
programs usually involve
elements from each of the
models.
Tannenbaum, 17 articles
The quantity of TB research decreased; Presented a comprehensive
Beard, &
however, the quality had improved.
model of team effectiveness
Salas
Qualitative
Most studies used multiple components
that continues to influence
(1992)
data
in their TB interventions.
research and theorizing
More researchers began using
in this field.
Years spanned:
behavioral and objective measures.
1980–1988 Presented evidence to cast doubt the
connection between process and
performance; TB is effective, but
for only perceptions and attitudes.
(continued)
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
191
Table 2 (continued)
Review
Study
Details
Summary of
Findings
Noteworthy
Features
Post-intervention strategies may be a
key mechanism to ensure the
long-term effectiveness of these
interventions.
Salas, Rozell, 11 articles
No significant effect of TB on
First known attempt to
Mullen, &
performance.
empirically summarize the
Driskell
Quantitative
A nonsignificant tendency for TB to
effectiveness of TB.
(1999)
data
result in lower performance when
Assessed a number of
measured objectively, but increase
moderators, including team
Years spanned:
performance when subjective
building component, team
1965–1990
measures were used.
size, training duration, and
The role clarification component was
type of performance measure
used (i.e., objective
more likely to increase performance.b
The effects of TB decreased as the size
versus subjective).
of the team increased.
The effects of TB decreased as the
duration of the intervention increased.
Note: TB = team building.
a. Many of the 36 studies reported evaluations of multiple dependent variables.
b. This result should be interpreted with caution as it was based on a small number of effect sizes (it’s difficult to determine, but likely only three or four effect sizes were used in this calculation).
Hypotheses
Is Team Building Effective?
We agree with Salas and colleagues (1999) that previous narrative
reviews have frequently expressed the benefits that can result from team
building (e.g., Buller, 1986; DeMeuse & Liebowitz, 1981; Sundstrom,
DeMeuse, & Futrell, 1990, Tannenbaum et al., 1992; Woodman &
Sherwood, 1980); however, there has been a lack of definitive, compelling
evidence concerning the positive effects of team building on team performance.
Specifically, previous qualitative reviews of the team-building domain have
concluded that evidence of an effect of team building on performance was
“inconclusive” (Buller, 1986), “unsubstantiated” (Woodman & Sherwood,
1980), “equivocal” (Tannenbaum et al., 1992), and “mixed” (Sundstrom
et al., 1990). Meta-analytic results from one study have suggested there is
no overall effect of team building on team performance (Salas et al., 1999).
However, there was support for the role-clarification component of team
building. Theoretically, one would assume that an intervention focused on
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
192
Small Group Research
Figure 1
Theoretical Model Depicting Study Hypotheses
improving team functioning would result in positive (as opposed to negative) outcomes. In addition, the moderate support suggested by several
narrative reviews (e.g., Tannenbaum et al.) leads us to predict a positive
overall effect of team building on team functioning. Thus, we present our
first hypothesis:
Hypothesis 1: Team building interventions will result in enhanced team outcomes.
The result from testing this hypothesis will provide for a baseline judgment concerning the efficacy of this particular form of team-development
intervention. The omnibus test will therefore allow for an overall assessment of the efficacy of team building, with all independent outcomes from
primary studies combined for this analysis.
Is Team Building More Effective for Some
Outcomes Than Others?
This framing question asks whether the combined set of team-building
interventions is shown to be more useful for improving certain team outcomes
than others. Specifically, does team building work better for cognitive
outcomes (e.g., declarative knowledge of teamwork competencies), team
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
193
member affective outcomes (e.g., trust, team potency), team processes
(e.g., coordination, communication), or team performance outcomes (e.g.,
volume of sales, productivity measures)? This division of team-building
outcomes is similar to the commonly discussed cognitive, affective, and
skill-based breakdown of general training outcomes (e.g., Kraiger, Ford, &
Salas, 1993). However, for the current research, skill-based outcomes are
further divided into two additional categories—team processes and more
performance-related or productivity-related outcomes.
The division and examination of team-building effectiveness based on
specific outcomes is intended to help clarify the often disparate results
seen in the literature on these interventions. For example, Woodman and
Sherwood’s (1980) qualitative review of team building concluded that it
was only useful for facilitating affective outcomes, not team performance.
Equally pejorative to the position that team building results in improved
team performance was the conclusion provided by Buller (1986), who upon
reviewing team-building studies conducted through 1980, suggested that
the relationship between team building and performance was inconclusive.
In yet another review that added to the opaque nature of the efficacy of team
building, DeMeuse and Liebowitz (1981) accurately noted that most of the
early research on team building relied almost exclusively on perceptual
ratings of the dependent variables being studied.
More recent reviewers of the efficacy of team building have reported
somewhat different (i.e., more positive) conclusions. For example, the integration of team-building research reported by Salas and colleagues (1999)
has served, in many ways, as the focal point for the current research. And,
although there was no significant overall effect of team building in their
research, there was a small, yet significant, tendency for team building to
increase performance when criteria were assessed with subjective measures.
In another investigation into the efficacy of team building, Tannenbaum and
colleagues (1992) reviewed team-building research conducted in the 1980s
and found support for these interventions, especially when the outcome of
interest was limited to team member perceptions or attitudes. Finally,
Svyantek, Goodman, Benz, and Gard (1999) found that team building positively impacted work-group productivity. Specifically, the largest impact of
team building was on productivity measures of cost-effectiveness, with a
smaller influence on quantity and quality measures.
In general, previous research that has investigated the effectiveness of
team building for improving specific outcomes has been equivocal at best,
especially considering performance or productivity measures. At the same
time, there has been perhaps the greatest support for the efficacy of team
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
194
Small Group Research
building for improving affective or attitudinal outcomes. Finally, to our
knowledge there have been no reviews citing the correlation between teambuilding interventions and improvements in cognitive outcomes.
The mixed results found in the literature concerning the effect of teambuilding interventions on team outcomes lead us to believe that moderators
may exist. Therefore, team building impacts certain outcomes more than
others. Based on the general characteristics of team building, (e.g., that
team building develops interpersonal relations, mutual trust, and open communication between team members), it is likely that team building will
have a greater effect on affective outcomes than any on other type of outcomes. As a process intervention, it makes theoretical sense that team
building would result in enhanced team member affective outcomes.
Finally, although the findings of more distal reviews of team building were
taken into consideration, a decision was made to place more credence on
recent research, which more often reported positive findings. These findings, along with a complete consideration of the expected benefits of team
building, led us to Hypotheses 2a and 2b:
Hypothesis 2a: Team building interventions will result in improved outcomes
across each of the four outcome types.
Hypothesis 2b: Team building will be most effective for improving affective
outcomes.
Does the Focus of Team Building Moderate Its Effectiveness?
It is certainly important to be able to identify whether team building
results in enhanced team functioning, but it is perhaps more informative to
know which forms or components of team building are most effective.
Advancements in theory in the last quarter century (e.g., Buller, 1986; Salas
et al., 1999; Salas, Priest, & DeRouin, 2005) have allowed for the parceling of team-building interventions into four distinct foci (see Table 1).
Unfortunately, the single existing empirical integration on the relative efficacy of different components of team building (i.e., Salas et al.) did not
provide encouraging results. Combining subjective and objective estimates,
there was a slight (nonsignificant) tendency for goal-setting (r = −.16),
interpersonal relations (r = −.06), and problem-solving (r = −.05) models
of team building to result in decreased performance. It was only the
role-clarification component that appeared to be effective for improving
performance (r = .76). However, upon examination of their data set, it
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
195
appears this positive finding for role clarification was supported by the
results from only three studies and three effect sizes. Thus, caution and further
investigation are warranted before concluding that the role-clarification
component of team building is superior to the other forms.
Nonetheless, because the role-clarification component of team building
is designed to relieve role stress as created by role ambiguity or role conflict, it is reasonable to anticipate that a significant improvement in team
functioning should result. In addition, the role-clarification component of
team building emphasizes communication among team members, and
thus it is likely that an increase in the level and quality of communication
between team members will impact their effectiveness. Unlike the other
forms of team building, improvements in role clarity and communication
are expected to produce more lasting benefits in terms of team functioning (as assessed through an analysis of the combined set of team outcomes).
Combining this theoretical rationale with the preliminary findings of
Salas and colleagues (1999) concerning the role-clarification component
of team building, it is our belief that a team-building intervention that
utilizes a role-clarification focus will provide the most benefit to team
functioning. At the same time, it is our view that any carefully thought out
team-building intervention should have at least some positive impact on
team members. Thus, we also expect that team building that focuses on
interpersonal relations, goal setting, or problem solving will also prove
useful, at least for the short-term benefits that are typically assessed.
Taking these dual considerations into perspective, we present our next set
of hypotheses:
Hypothesis 3a: Each of the four components of team building will demonstrate a moderate level of effectiveness for improving team functioning.
Hypothesis 3b: The role clarification component of team building will be
most effective for improving team functioning.
Is the Effectiveness of Team Building
Moderated by Team Size?
This research will also investigate whether the effectiveness of team
building is moderated by the size of the team. At a basic level, “the resources
available on a team result from how many people are on it” (Hambrick &
D’Aveni, 1992, p. 1449). Sundstrom, McIntyre, Halfhill, and Richards’s
(2000) review of 83 field studies and experiments conducted with work
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
196
Small Group Research
groups suggested that the average group size was 11 members. Moreover,
Halebian and Finkelstein (1993) have suggested that team size is synonymous with cognitive capability. Providing support for this assertion, Bantel
and Jackson (1989) found that larger teams generally have a greater reservoir
of cognitive resources than smaller teams. There is little doubt that there are
benefits to having medium- to large-sized teams available to perform work
tasks rather than smaller teams. However, research has also found that larger
teams can facilitate the enactment of other, less desirable, group phenomenon, including the participation leadership effect, the in-group bias effect, the
cohesiveness performance effect, and the well-known groupthink effect (cf.
Mullen, Anthony, Salas, & Driskell, 1994; Mullen, Brown, & Smith, 1992;
Mullen & Copper, 1994; Mullen, Salas, & Driskell, 1989).
From other research, we know that information pooling is critical to teambased decision making. That is, the size of a team may impact the effectiveness
with which a team pools common and unique information. For example, Stasser,
Vaughan, and Stewart (2000) discussed the tendency for group members to discuss shared information rather than the unique information that is held by team
members—an issue that becomes more problematic as teams increase in size.
This lack of attention to unique information held by individual team members
can lead to mal-informed decisions and occasionally even detrimental results.
Salas and colleagues (1999) presented the only existing research integration
that has assessed the relative efficacy of team building for teams of different
sizes. They found that the effects of team building on performance decreased
as a function of the size of the team, both in objective measures and in subjective measures of performance. These authors concluded that any positive effect
of team building is most likely to prevail only in small teams. Similarly, others
have argued that as group size increases, members’ liking for the group (Indik,
1965) and performance (Mullen, 1987) tend to decrease. Taking these findings
into consideration, the current research examined team size as a moderator of
the effectiveness of team-building efforts on team performance. In this study,
we seek to replicate Salas and colleagues’ findings. However, rather than simply correlating the database of effect sizes with their associated team sizes, the
current research will examine the efficacy of team building for three distinct
subgroup classifications of team size: small teams (i.e., less than 5 members);
medium-sized teams (i.e., 5 to 10 members); and large teams (i.e., greater than
10 members). In addition, although there is some reason to expect that larger
teams have an increased cognitive capacity with which to perform tasks (e.g.,
Bantel & Jackson, 1989), the other negative issues often associated with
increased group size are expected to be the overwhelming influences on the
performance of teams. Therefore, it is reasonable to suggest that larger teams
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
197
are already performing at a lower level than medium or small teams, and would
therefore exhibit enhanced benefits from team-building interventions. Stated
differently, as teams increase in size, it is more likely they will show substantial benefits from team building. This expectation, although supported by
theory, is in direct contradiction to the findings reported by Salas and colleagues. The following hypothesis is proposed to investigate this assertion.
Hypothesis 4: Large teams will show greater benefits from team building
than small- or medium-sized teams.
Method
Literature Search
A comprehensive literature review was conducted to identify published
and unpublished studies relevant to the effects of team building on the four
team outcomes. As a starting point for our literature review, we conducted
an online search of the Defense Technical Information Center (DTIC),
Google Scholar ©, and the most common academic search databases
(e.g., PsycINFO). The particular keywords that were used included, but were
not limited to, the following terms: team building, team development, team
goal setting, interpersonal relations, problem solving, role clarification, group
building, and group development. In addition to these search techniques, key
articles relevant to team building were inspected by hand for additional,
potentially useful primary studies (e.g., Buller, 1986; DeMeuse & Liebowitz,
1981; Salas et al., 1999; Tannenbaum et al., 1992; Woodman & Sherwood,
1980). This ancestry approach was later extended to each of the articles that
were included in the database in an effort to ensure that no fugitive studies
had been overlooked. In the end, the literature search process resulted in 103
articles being identified for potential inclusion in the database.
Criteria for inclusion. Specific limits were placed on the search to include
articles published from 1950 to 2007. This date range was selected because
previous reviews had not uncovered any published or unpublished teambuilding evaluations prior to 1950. Moreover, studies were included only if
they involved adult, human, and nonclinical populations, and focused on
teams and not groups. Finally, studies had to report data sufficient to calculate an effect size (r) assessing the relationship between team-building interventions and outcomes.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
198
Small Group Research
Coding Procedure
Following the literature review, three authors independently coded each of
the relevant studies on 14 categories: (a) nature of the organization and participant sample; (b) team type; (c) number of teams; (d) average team size;
(e) predictor reliability; (f) level of analysis of predictor; (g) criterion reliability; (h) level of analysis of the criterion; (i) criterion report type(s); (j) criterion description(s); (k) focus/component of team building (i.e., goal setting,
interpersonal relations, problem solving, role clarification); (l) effect size(s);
(m) study design type; and (n) recommendation for inclusion. Concerning the
level of analysis coding, the initial database consisted of 69 effect sizes—60
at the team level and 9 at the individual level. However, mixing levels of
analysis is not recommended for research integrations for numerous reasons
(e.g., Gully, Devine, & Whitney, 1995; Hunter & Schmidt, 1990). Therefore,
only the team-level outcomes were analyzed in the current research, reducing
our database to the 60 team-level correlations.
Another point concerning the coding process that bears clarification is the
assessment of the focus or component of team building. It was expected that
many of the interventions being examined would consist of multiple components of team building. Thus, coders were instructed to allocate a total of
100 percentage points to each of the four components of team building. For
example, if a coder believed an intervention consisted of equal parts role
clarification and interpersonal relations, each was coded with 50 percentage
points, whereas problem solving and goal setting were given a 0 for this coding category. For this process, the coders were instructed to closely examine
the effort, intensity, and primary focus of the intervention. They were
instructed further that they would need to justify the weights they assigned
during a consensus discussion of the final articles included in the database. In
this discussion, the three coders came to a consensus concerning the allocation of percentage points to each team-building intervention.
Rater reliability. Articles were coded by one of three of the authors. To
aid in the consistency of coding, the three authors met early in the process
to discuss the 14 pieces of information that were to be extracted from each of
the 103 articles. Moreover, the coders had previously pilot tested the coding
scheme, and possessed satisfactory level of expertise in the substantive
areas being coded. To estimate the reliability with which the coders were
evaluating the primary studies, three authors independently coded the same
set of 20 articles (the reliability sample, representing approximately 19% of
the total 103 articles selected for coding) to determine interrater reliability
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
199
by calculating intraclass correlation coefficients (ICCs; Nunnally, 1978;
Shrout & Fleiss, 1979). The coders then each independently coded approximately one third of the remaining articles.
During the coding process, the decision to include or exclude the article for
the final analyses was of primary concern. Upon inspection of the 20 studies
coded by each of the three coders, it was revealed that all three coders were in
agreement 85% of the time with their global evaluation to include (or not to
include) particular studies. The ICC for this assessment was satisfactory
(3, k) = .85. Importantly, for the three studies for which there was not complete
initial agreement the authors met to discuss the discrepancies and a consensus
was reached. The ICCs calculated for seven of the categories ultimately used
in either main or exploratory analyses ranged from ICC (3, k) = .85 to ICC (3,
k) = 1.0. In addition, the agreement among the three coders across all seven
categories resulted in an ICC (3, k) = .96. The remaining six categories that
were utilized for the ICC calculations included: team type, number of teams
in the primary studies under investigation, average team size, focus/component of the team-building intervention, effect size estimate, and study design.
The ICCs for these categories were .94, .94, .97, .94, .91, and 1.0, respectively.
In conclusion, the consistency and agreement among the raters was very good.
Meta-Analysis Procedure
The software for the Hunter-Schmidt meta-analysis methods was used to
analyze the data (Schmidt & Le, 2005). This program provides output that
includes, but is not limited to: (a) the mean true score correlation, (b) the variance and standard deviation of true score correlations, (c) credibility intervals,
(d) estimates of the variance and standard deviation in observed correlations
due to artifacts, and (e) the percentage of variance attributable to observed correlations after the removal of artifacts. In addition, the software employs a random effects model to combine effect sizes from primary studies. Random
effects meta-analysis models allow the true effect sizes to vary, in contrast with
fixed-effects models that assume the true effect sizes have fixed values.
Moreover, the random effects model is considered to be “more realistic than
the fixed-effect model on the majority of occasions” (Field, 2001, p. 162).
Effect-Size Calculations
Before calculating a meta-analytic estimate of the relationship between
variables, effect sizes culled from primary studies must be prepared for
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
200
Small Group Research
entry into the database. During this process, effect sizes from primary studies were converted to the common metric r (correlation). Thus, when necessary, primary study effect sizes reported as other statistics (e.g., t, F, d, χ2,
or Z) were transformed using the formulas found in Hunter and Schmidt
(2004). Once placed on this common metric of effect size, the results of
independent tests found in primary studies can be combined and assessed
for fit with predicted hypotheses in the meta-analysis. However, it is
common knowledge that when a study contains multiple effect sizes, they
are stochastically dependent (Shadish, Cook, & Campbell, 2002). These
dependencies violate the statistical assumption of independent effect
sizes. The recommended solution is to average or combine effect sizes
from different measures of the same sample within a single study prior to
combining results from multiple studies.
Corrections for Unreliability and Range Restriction
It is common in meta-analyses to make attempts to correct obtained
reliability coefficients for measures of the predictor, criterion, or both (e.g.,
Hunter & Schmidt, 1990; Johnson, Mullen, & Salas, 1995). Unfortunately,
original studies often fail to report all of the auxiliary information necessary
to perform corrections for study artifacts. Such was the case with the current database. As a result, artifact distribution meta-analysis was employed,
rather than correcting each effect size individually for unreliability.
Authors of meta-analytic integrations also make attempts to correct for
the effects of direct or indirect range restriction. Performing corrections for
range restriction will generally result in a combined estimate that is more
accurate than had no corrections been performed at all (e.g., Hunter,
Schmidt, & Le, 2006). However, in the current investigation, primary
studies did not report information on restricted and unrestricted samples
that would be necessary to perform corrections for range restriction. We
therefore cannot correct for range restriction in the current meta-analysis.
Consequently, our effect-size estimates may be conservative, in the sense
that they may underestimate the true effect of team building if the biasing
effects of range restrictions could be corrected.
Weighting
Primary study effect sizes included in the current series of meta-analytic
integrations were weighted by their sample sizes. It is presumed that effect
sizes obtained from studies with larger sample sizes are more stable (i.e.,
accurate) than those culled from studies with small sample sizes. Although
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
201
it is important to remember that large sample studies are not necessarily
more valid, recent simulation studies have determined that a very accurate
combined effect size can be obtained through procedures used to weight
primary studies by sample size (e.g., Field, 2005).
Results
Description of the Database
A total of 60 correlations were obtained from 20 studies. These 60 effect
sizes represented 1,562 teams, with a median team size of approximately 9
members. Although these 60 effect sizes were not all from independent
samples, every subgroup analysis that was performed included only independent samples. Otherwise, for the overall assessment of the influence of
team building on the combined set of team outcomes, there were 26 independent samples (see Table 3). Of the studies included in the meta-analytic
database, 14 were published and 6 were unpublished. Table 3 provides a
description of the key information derived from each primary study.
Publication/Availability Bias Detection
When performing any meta-analysis it is difficult to determine whether
all relevant studies have been located. Existing empirical evaluations that
are elusive to locate and retrieve are commonly referred to as fugitive
literature (Rosenthal, 1994). To the extent that this fugitive literature represents a substantial proportion of the conducted evaluations in any one area,
there is a possibility of publication or availability bias. The file drawer
problem in meta-analyses presents itself when there is a concern that the
studies that find their way into publication are simply the ones that show
significant results, whereas evaluations with nonsignificant findings are relegated to file drawers (Rosenthal, 1979). To alleviate this concern,
Rosenthal’s (1979) file drawer analysis can be used to provide an estimate
of the number of unpublished studies relegated to file drawers. This analysis designates the number of these articles, which average null results, that
would be required to bring the significance level for a set of studies down
to the just-significant level (Hunter & Schmidt, 2004). However, it has been
demonstrated that the widely used fail-safe file drawer analysis is essentially irrelevant and can result in incorrect estimates of the size of the file
drawer (Scargle, 2000). Instead, funnel plots can serve as an alternative to
file drawer analysis for the purpose of detecting the possibility of bias.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
202
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
1981
1981
1981
1973
1986
1986
1986
1986
1986
1986
1995
1995
1995
1995
1995
1995
1993
1993
1993
Boss & McConkie
Boss & McConkie
Bragg & Andrews
Buller & Bell
Buller & Bell
Buller & Bell
Buller & Bell
Buller & Bell
Buller & Bell
Bushe & Coetzer
Bushe & Coetzer
Bushe & Coetzer
Bushe & Coetzer
Bushe & Coetzer
Bushe & Coetzer
Cohen
Cohen
Cohen
Year
Boss & McConkie
Study
Author(s)
SGPP
PPWC
PPWC
SGPP
SGPP
SGPP
SGPP
SGPP
SGPP
PPWC
PPWC
PPWC
PPWC
PPWC
PPWC
SGPP
SGPP
SGPP
SGPP
Design
Type
0.13
0.17
0.83
0.86
0.85
0.70
0.59
0.57
0.66
0.40
0.36
0.28
0.33
0.92
0.95
0.16
0.89
0.85
0.91
Effect
Size, r
0.00
0.00
0.00
0.00
1.00
0.50
1.00
0.00
0.50
0.00
0.50
0.00
0.50
0.00
0.50
0.00
0.00
0.00
0.00
GS
1.00
1.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.00
1.00
1.00
1.00
IR
0.00
0.00
1.00
1.00
0.00
0.50
0.00
1.00
0.50
1.00
0.00
1.00
0.00
1.00
0.00
0.00
0.00
0.00
0.00
PS
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.50
0.00
0.50
0.00
0.50
0.00
0.00
0.00
0.00
RC
Table 3
Meta-Analytic Database
Executive/
Management
Executive/
management
Executive/
management
Production
Production
Production
Production
Production
Production
Production
Project
Project
Project
Project
Project
Project
Action/
performing
Action/
performing
Action/
performing
Team
Type
4
4
3
12
8
12
8
12
12
16
16
16
16
16
16
4
2
2
2
Number
of Teams
16.67
16.67
32.00
16.00
16.00
16.00
16.00
16.00
16.00
4.00
4.00
4.00
4.00
4.00
4.00
16.67
10.00
10.00
10.00
Average
Team Size
(continued)
Process
Affective
Performance
Process
Process
Process
Performance
Performance
Performance
Affective
Affective
Process
Process
Performance
Performance
Affective
Performance
Process
Affective
Criterion
Description
203
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
PPWC
PPWC
PPWC
PPWC
PPWC
PPWC
PPWC
PPWC
PPWC
PPWC
SGPP
PPWC
1998
1998
1986
1986
1967
1967
1967
2001
2001
1979
1979
2002
2002
2002
1983
1983
Dionne
Dionne
Eden
Eden
Friedlander
Friedlander
Friedlander
Gibson
Gibson
Howard
Howard
Huang, Wei,
Watson, & Tan
Huang, Wei,
Watson, & Tan
Huang, Wei,
Watson, & Tan
Hughes, Rosenbach,
& Clover
Hughes, Rosenbach,
& Clover
PPWC
SGPP
SGPP
PPWC
PPWC
1993
Cohen
Design
Type
Year
Study
Author(s)
0.23
0.50
0.68
0.97
0.00
0.16
0.21
0.24
−0.12
0.25
0.40
0.97
0.36
0.20
0.05
0.13
0.18
Effect
Size, r
0.50
0.50
1.00
1.00
0.00
0.00
0.00
1.00
1.00
0.00
0.00
1.00
0.33
0.00
0.00
0.33
0.00
GS
0.50
0.50
0.00
0.00
0.50
0.50
0.50
0.00
0.00
1.00
1.00
0.00
0.33
1.00
1.00
0.33
1.00
IR
0.00
0.00
0.00
0.00
0.50
0.50
0.50
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
PS
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.33
0.00
0.00
0.33
0.00
RC
Table 3 (continued)
Project
Project
Production
Production
Action/
performing
Project
Project
Action/
performing
Action/
performing
Project
Project
Project
Service
Service
Service
Service
Production
Team
Type
2
2
48
48
12
12
12
71
71
3
3
48
16
54
54
16
4
Number
of Teams
68.00
74.00
5.00
5.00
10.00
10.00
10.00
5.00
5.00
9.67
9.67
5.00
30.00
4.50
4.50
30.00
16.67
Average
Team Size
(continued)
Affective
Affective
Performance
Process
Affective
Process
Performance
Affective
Performance
Affective
Performance
Affective
Process
Process
Performance
Performance
Process
Criterion
Description
204
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
SGPP
SGPP
SGPP
SGPP
SGPP
SGPP
SGPP
SGPP
PPWC
SGPP
SGPP
1975
1975
1975
1975
1994
1997
1997
1986
1986
1986
1980
1980
Miller
Mitchell
Mitchell
Mitchell
Morrison &
Sturges
Morrison &
Sturges
SGPP
PPWC
SGPP
Design
Type
1983
1983
Year
Hughes, Rosenbach,
& Clover
Hughes, Rosenbach,
& Clover
Kimberley &
Nielsen
Kimberley &
Nielsen
Kimberley &
Nielsen
Kimberley &
Nielsen
Longenecker,
Scazzero, &
Stansfield
Miller
Study
Author(s)
0.47
0.72
0.25
0.78
0.67
0.06
0.18
0.83
0.19
0.14
0.38
0.48
0.03
0.52
Effect
Size, r
0.00
0.00
0.00
0.00
0.00
0.10
0.10
1.00
0.00
0.00
0.00
0.00
0.50
0.50
GS
0.50
1.00
1.00
1.00
.50
0.10
0.10
0.00
0.00
0.00
0.00
0.00
0.50
0.50
IR
0.00
0.00
0.00
0.00
0.00
0.70
0.70
0.00
1.00
1.00
1.00
1.00
0.00
0.00
PS
0.50
0.00
0.00
0.00
0.50
0.10
0.10
0.00
0.00
0.00
0.00
0.00
0.00
0.00
RC
Table 3 (continued)
Action/
performing
Action/
performing
Project
Project
Project
Executive/
management
Executive/
management
Production
Production
Production
Production
Production
Project
Project
Team
Type
2
12
12
9
2
16
16
2
90
90
180
180
2
2
Number
of Teams
12.00
4.50
4.50
4.50
12.00
7.38
7.38
45.00
10.00
10.00
10.00
10.00
68.00
74.00
Average
Team Size
(continued)
Process
Affective
Affective
Affective
Affective
Process
Affective
Performance
Performance
Performance
Process
Affective
Cognitive
Cognitive
Criterion
Description
205
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
SGPP
SGPP
SGPP
SGPP
PPWC
PPWC
PPWC
PPWC
PPWC
PPWC
1983
1990
1990
1990
1990
1980
1980
1980
1980
Design
Type
1980
Year
0.15
0.23
0.10
0.39
0.24
−0.22
0.39
−0.46
0.10
0.47
Effect
Size, r
0.50
0.50
0.50
0.00
0.00
0.00
0.00
0.00
0.50
0.00
GS
0.00
0.00
0.00
0.25
0.00
1.00
0.00
1.00
0.00
0.50
IR
0.50
0.50
0.50
0.50
1.00
0.00
1.00
0.00
0.50
0.00
PS
0.00
0.00
0.00
0.25
0.00
0.00
0.00
0.00
0.00
0.50
RC
Project
Project
Project
Executive/
management
Service
Service
Service
Service
Service
Project
Team
Type
67
67
67
4
4
4
4
4
67
2
Number
of Teams
3.50
3.50
3.50
6.00
6.00
4.50
5.50
4.75
3.50
12.00
Average
Team Size
Performance
Process
Affective
Affective
Process
Process
Process
Process
Cognitive
Performance
Criterion
Description
Note: GS = goal-setting team building; IR = interpersonal relations team building; PS = problem-solving team building; RC = role-clarification team building; PPWC =
pre–post with control group comparison; SGPP = single-group pre–post comparison.
Morrison &
Sturges
Wegenast
Wexler
Wexler
Wexler
Wexler
Woodman &
Sherwood
Woodman &
Sherwood
Woodman &
Sherwood
Woodman &
Sherwood
Study
Author(s)
Table 3 (continued)
206
Small Group Research
Funnel plot. The funnel plot (Light & Pillemer, 1984) is a simple visual
tool (i.e., scatterplot) for detecting the presence of publication or other
availability bias in meta-analysis. To apply this technique, effect sizes are
graphed on the horizontal axis, whereas study sample sizes (N) are graphed
on the vertical axis. The idea is that, in the absence of bias, the results of
small sample studies will scatter widely at the bottom of the graph, with the
spread narrowing for larger N studies; thus taking the form of an inverted
funnel (Hunter & Schmidt, 2004). If publication bias is present in the data,
the figure will be asymmetrical and often truncated in the lower left-hand
portion of the scatterplot. As seen in Figure 2, the data from primary studies included in the current meta-analysis form a somewhat regular funnel.
Thus, it appears that the small sample studies are spread rather evenly
across the effect-size continuum, making the possibility of publication or
availability bias less of a concern in this research.
Meta-Analytic Results
The meta-analytic results for the hypotheses investigated in this research
are presented in Tables 4 through 6. These tables show a number of pieces
of information, including: the number of teams in each analysis (N); the
number of independent effect sizes (correlations) in each analysis (k); the
mean weighted observed correlation ( r̄); the 80% confidence interval for
that correlation; the estimated true score correlation (ρ); the standard
deviation of this true score correlation (SDρ); the 80% credibility interval
(10% CV and 90% CV); and the percentage of variance accounted for by
statistical artifacts. Confidence and credibility intervals are useful as aids in
providing the best estimate of the true nature of the relationships between
two variables (Whitener, 1990). Confidence intervals are applied to observed
scores, center on a single mean score, and reflect the effects of sampling
error. Credibility intervals, on the other hand, are particularly meaningful
because they take into account information about the distribution of effect
sizes after other research artifacts have been taken out. Credibility intervals
can also be useful for determining whether moderators are operating
(Whitener, 1990). Similarly, the percentage of variance estimate provides
information concerning the variance in observed correlations that is the
result of statistical artifacts. In general, the higher the percentage, the more
certain we can be that additional moderators are not operating. The following
sections more closely examine the results for the individual hypotheses that
were put forward in this research.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
207
Figure 2
Funnel Plot for Detecting the Possibility of
Publication or Availability Bias
100
Number of Teams
80
60
40
20
0
–0.500
–0.250
0.000
0.250
0.500
0.750
1.000
Effect Size
Team-building effectiveness. The omnibus test representing all teambuilding interventions and outcomes resulted in a significant tendency for
these interventions to improve team outcomes. Specifically, the mean true
score correlation of .31 represents a moderate effect (10% CV = .11; 90%
CV = .52) and provided support for Hypothesis 1. This analysis included 26
independent effect sizes and was based on a total sample size of 579 teams.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
208
Small Group Research
Table 4
Analysis of the Effectiveness of Team Building
Based Upon Outcome Type
Outcome
Type
Cognitive
Affective
Process
Performance
All
outcomes
All
outcomesa
N
k
r̄
CIr
10%
CIr
90%
ρ
SDρb
10%
CV
90%
CV
% Var.
Acct.c
71
482
485
524
579
3
19
20
18
26
.11
.41
.39
.25
.28
.06
.34
.31
.16
.22
.17
.49
.46
.33
.29
.13
.44
.44
.26
.31
.00
.18
.20
.23
.16
.13
.21
.18
−.03
.11
.13
.66
.69
.55
.52
1,394.24
55.42
54.04
45.11
69.35
1,562
60
.34
.29
.38
.37
.21
.10
.64
50.64
Note: k = number of correlations coefficients on which each distribution was based; r̄ = mean
observed correlation; CIr 10% = lower bound of the confidence interval for observed r; CIr
90% = upper bound of the confidence interval for observed r; ρ = estimated true correlation
between the predictor construct and the relevant criterion (fully corrected for measurement
error in both the predictor and the criterion); SDρ = estimated standard deviation of the true
correlation; 10%CV = lower bound of the credibility interval for each distribution; 90% CV =
upper bound of the credibility interval for each distribution; % var. acct. = percentage of
observed variance accounted for by statistical artifacts.
a. This result represents the entire database and patently violates the assumption of independent effect sizes as many studies contributed more than one effect size. It is shown here for
illustrative purposes only.
b. The SDρ being zero indicates that the real variance of the true correlation is zero—there is
only one value of the true correlation underlying all the studies. It is a consequence of the
percentage of variance accounted for estimate being greater than 100%. This result also indicates that there should be no additional moderators operating for this analysis.
c. The percentage of variance explained estimate being greater than the theoretical maximum
value of 100% indicates that sampling error and other study artifacts explain all of the
observed variation in the effect sizes (correlations) across studies. The estimated value is
greater than 100% because of second-order sampling error.
Here, cognitive, affective, process, and performance outcomes were averaged for each primary study before figuring into the combined estimate. The
only exception was for six of the primary studies that reported the results of
team-building interventions for separate samples within the same study. For
these instances only, there were multiple correlations from single studies
that contributed to the combined estimate. However, care was taken to
ensure the assumption of independent effect sizes was not violated.
Investigation of separate outcomes. From the overall database of 60 effect
sizes, this research also assessed the impact of team building on distinct
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
209
Table 5
Analysis of the Effectiveness of Team Building Based Upon
Team-Building Component
Team-Building
Component
Goal setting
Interpersonal
relations
Problem solving
Role clarification
N
k
r̄
CIr
10%
CIr
90%
ρ
SDρa
10%
CV
90%
CV
% Var.
Acct.b
258
140
10
13
.34
.23
.21
.15
.47
.31
.37
.26
.27
.00
.02
.26
.71
.26
37.65
233.50
326
54
11
5
.23
.32
.16
.22
.29
.42
.24
.35
.00
.00
.24
.35
.24
.35
114.27
322.06
Note. k = number of correlations coefficients on which each distribution was based; r̄ = mean
observed correlation; CIr 10% = lower bound of the confidence interval for observed r; CIr
90% = upper bound of the confidence interval for observed r; ρ = estimated true correlation
between the predictor construct and the relevant criterion (fully corrected for measurement
error in both the predictor and the criterion); SDρ = estimated standard deviation of the true
correlation; 10% CV = lower bound of the credibility interval for each distribution; 90% CV
= upper bound of the credibility interval for each distribution; % var. acct. = percentage of
observed variance accounted for by statistical artifacts.
a. The SDρ being zero indicates that the real variance of the true correlation is zero—there is
only one value of the true correlation underlying all the studies. It is a consequence of the percentage of variance accounted for estimate being greater than 100%. This result also indicates
that there should be no additional moderators operating for this analysis.
b. The percentage of variance explained estimate being greater than the theoretical maximum
value of 100% indicates that sampling error and other study artifacts explain all of the
observed variation in the effect sizes (correlations) across studies. The estimated value is
greater than 100% because of second-order sampling error.
outcomes. Specifically, we evaluated the impact team building had on cognitive, affective, process, and performance outcomes. The number of effect
sizes for these subgroup analyses ranged from 3 to 20. Table 4 presents the
results of these investigations.
In evaluating the impact of team building on cognitive outcomes, three
effect sizes were analyzed (N = 71 teams). The estimated true score correlation (ρ) for the relationship between team building and improvements in
cognitive outcomes was .13. Although it appears that team building has a
negligible impact on these outcomes, this result should be interpreted with
extreme caution since there were only three effect sizes available to combine. For the analysis of the influence of team building on affective outcomes, the estimated true score correlation was .44 (10% CV = .21; 90%
CV = .66). This result represented the accumulation of data from 19 effect
sizes and a total of 482 teams. Concerning process outcomes, the results
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
210
Small Group Research
Table 6
Analysis of the Effectiveness of Team Building Based Upon Team Size
Team
Size
Small
Medium
Large
N
k
r̄
CIr
10%
CIr
90%
ρ
SDρa
10%
CV
90%
CV
% Var.
Acct.b
178
340
61
7
10
9
.26
.25
.54
.15
.14
.44
.37
.36
.64
.28
.27
.66
.12
.21
.00
.12
.00
.66
.44
.54
.66
74.22
45.72
197.54
Note: k = number of correlations coefficients on which each distribution was based; r̄ = mean
observed correlation; CIr 10% = lower bound of the confidence interval for observed r; CIr
90% = upper bound of the confidence interval for observed r; ρ = estimated true correlation
between the predictor construct and the relevant criterion (fully corrected for measurement
error in both the predictor and criterion); SDρ = estimated standard deviation of the true correlation; 10% CV = lower bound of the credibility interval for each distribution; 90% CV =
upper bound of the credibility interval for each distribution; % var. acct. = percentage of
observed variance accounted for by statistical artifacts.
a. The SDρ being zero indicates that the real variance of the true correlation is zero—there is
only one value of the true correlation underlying all the studies. It is a consequence of the percentage of variance accounted for estimate being greater than 100%. This result also indicates
that there should be no additional moderators operating for this analysis.
b. The percentage of variance explained estimate being greater than the theoretical maximum
value of 100% indicates that sampling error and other study artifacts explain all of the
observed variation in the effect sizes (correlations) across studies. The estimated value is
greater than 100% because of second-order sampling error.
suggested an estimated true score correlation of .44 (k = 20 correlations;
N = 485 teams). The 80% credibility interval for this finding ranged from
.18 to .69. Finally, for the analysis on performance outcomes, the results
indicated an estimated true score correlation of .26 (k = 18; N = 52).
All told, the results of the analyses focused on the influence of team
building on separate outcomes provided moderate support for Hypothesis 2a.
Specifically, team building appeared to be effective for improving each of
the four outcomes. However, it should be pointed out that the credibility
interval for performance outcomes barely included 0 (10% CV = −.03; 90%
CV = .55). At the same time, Hypothesis 2b received only partial support.
Although it was posited that team building would be most effective for
improving affective outcomes, the results from this study suggested that
team building was slightly more effective for improving process outcomes
(ρ = .439 vs. .437). However, given the considerable overlap in both confidence and credibility intervals for these analyses (see Table 4), it is difficult
to determine with any certainty which outcome type is most greatly affected
by team-building interventions. Moreover, team building did appear to be
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
211
more effective for improving affective outcomes than either cognitive or
performance outcomes (ρs = .13 and .26, respectively).
Focus of team building. A total of 39 correlations were meta-analyzed to
assess the differential impact of various team-building components. For this
analysis, the interventions described in primary studies were closely scrutinized to determine which component(s) of team building was the major
focus. In some instances it was determined that the intervention had a
multiple focus on two or more separate components. Therefore, in accumulating the study results, correlations from primary studies were sometimes included in multiple categories for the purposes of analysis so that
there was some degree of overlap regarding the analysis of team-building
components. Table 5 presents the results of these analyses.
Ten effect sizes, representing 258 teams, were meta-analyzed for the
investigation into the efficacy of the goal-setting component of team building.
The estimated true score correlation was .37, indicating a moderate effect on
the combined set of team outcomes. The remaining team-building components, namely, interpersonal relations, problem solving, and role clarification,
also resulted in moderate effect sizes. The estimated true mean score correlations were .26 (k = 13, N = 140), .24 (k = 11, N = 326), and .35 (k = 5,
N = 54), respectively. The credibility intervals for these analyses were also
calculated (see Table 5). All calculated intervals of the true mean score correlation did not include 0 (the 10% CV > .00 for all cases), thus indicating
that the true mean score correlations are positive in most of their populations.
In other words, the effects of team building based on all the components
examined generalize across most situations and settings. The results of this
analysis confirmed Hypothesis 3a, as each of the team-building components
resulted in improved team functioning. However, there was only partial support for Hypothesis 3b. Specifically, though the role-clarification component
of team building (ρ = .35) appeared to be superior to either the interpersonal
relations (ρ = .26) or problem-solving (ρ = .24) components, the goal-setting
component (ρ = .37) appeared to work best of all.
Team size. Twenty-six effect sizes were used to assess the potential moderating influence of team size on the ability of team building to improve
team functioning. The results for this analysis are presented in Table 6.
Effect sizes were categorized into three groups based on the average team
size. Effect sizes grouped into the small team category included those that
averaged less than 5 team members; the medium size category included
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
212
Small Group Research
those effect sizes based on an average team size of 5 to 10 members; and
the large team category included those effect sizes based on an average
team size of more than 10 members.
For small teams, a meta-analysis of 7 effect sizes, representing 178
teams, indicated a mean true score correlation of .28. For medium-sized
teams of 5 to 10 members, an analysis of 10 effect sizes representing 340
teams resulted in a mean true score correlation of .27. Thus, for both smalland medium-sized teams there was a moderate effect of team building on
team functioning. However, the results of a meta-analysis of team building
for large teams, which consisted of 9 effect sizes (N = 61), suggested that
the greatest impact of team building was upon teams that are large in size
(ρ = .66). This rather large effect is in direct support of Hypothesis 4, which
posited that the influence of team building would be most profound in large
teams, as compared to small- or medium-sized teams.
Discussion
The present study was conducted to answer the question of whether
team building works. The results are encouraging—they are suggestive of
the idea that team building does improve team outcomes. Specifically,
process and affective outcomes were most improved by team-building
interventions. Moreover, all the components (i.e., role clarification, goal
setting, interpersonal relations, and problem solving) of team building had
a moderate effect on outcomes but the goal-setting and role-clarification
components had the largest effect. Although teams of all sizes benefited
from team building, large teams appeared to benefit the most.
Despite the use of sophisticated meta-analytic techniques that were
employed to assist in the examination of the four hypotheses that served as
the focus for this research, our ability to definitively address the framing
questions put forth and test for the hypothesized effects was dependent on
the research available to us. In some instances, small subgroup effect sizes
precluded fine-tuned assessment and the derivation of definitive conclusions. However, the findings from the current research were still instructive
in many ways. As noted, the results from this study suggest that we are
beginning to find some positive empirical support for the effectiveness of
these commonly applied team-development strategies.
What else have we learned? In addition to finding support for the effectiveness of team building in general, it was deemed equally important to
understand whether the effectiveness of team building was moderated by
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
213
the specific outcomes that are targeted. Our second hypothesis posited that
team-building interventions would result in improved outcomes across each
of the four outcome types. This hypothesis was supported, with team building shown to be effective for each of the outcome types examined. Of the
four outcomes of interest, the results for cognitive outcomes were found to
be the least robust. However, given the exceptionally small number of correlation coefficients that contributed to this subgroup analysis (i.e., k = 3),
this finding may be properly viewed with considerable skepticism. The second part of this hypothesis was based upon prior theory and research in this
area, and posited that team building would be most effective for improving
affective outcomes. The findings indicated that team building is indeed
highly effective for improving team member affective outcomes, but may
be just as useful for facilitating improvements in team processes. Thus, it
remains an open question as to which outcome is most greatly affected by
team building—team processes or team member affective outcomes.
Our next research inquiry led us to investigate whether the focus of team
building moderated its effectiveness. First, it was found that each of the
team-building components were useful for enhancing team functioning,
with the estimated true score correlations ranging from .24 to .37. Similar
to the Salas and colleagues’ (1999) meta-analysis, our data suggested that
the role clarification component had the most impact on team outcomes.
However, in contrast to Salas and colleagues, it appeared that the goalsetting component of team building was equally and perhaps more effective
for improving team functioning as the role-clarification component. The
results from this research also suggested that the interpersonal relations and
problem-solving components were less effective.
This research also examined whether team building is more effective for
larger teams than smaller teams. Specifically, it was hypothesized that the
effect sizes for small- and medium-sized teams would be weaker than those
observed from team-building interventions conducted with large teams.
Our prediction was supported—there was a substantial estimated true score
correlation found for large teams and more modest results with teams that
were classified as small or medium. One possible explanation for this result
is the likely pre-intervention state of teams of various sizes. Although larger
teams generally have a greater reservoir of available cognitive resources
(Bantel & Jackson, 1989), it is also suggested that they are plagued by
problems such as groupthink, the participation leadership effect, and the
cohesiveness performance effect (e.g., Mullen et al., 1989). Given all the
potential problems that can accompany larger teams, it is feasible that
larger teams begin in a more negative state than smaller teams. Thus, there
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
214
Small Group Research
may be less room for improvements from team building to manifest themselves with teams of smaller size, which are typically more cohesive and
less likely to experience these problems.
In conclusion, this research extends the team-development literature by providing an empirical assessment of the relationship between team building interventions and outcomes. In addition, a number of potential moderators were also
investigated. By extending the meta-analysis by Salas and colleagues (1999),
more direction can be given to researchers who seek to further investigate specific questions regarding the effectiveness of team building. However, the fact
that our findings were not that consistent with those provided by Salas and colleagues deserves a brief explanation. Specifically, how should we interpret the
present findings in light of the contrasting findings of Salas and colleagues?
Importantly, our database included approximately twice as many articles
(20 vs. 11) and nearly four times as many correlations (60 vs. 16) as the previous effort. In addition, although their study was published in 1999, the last
included article in their database was published in 1990. In contrast, our database included seven sources put forward since 1990. The increased number of
effect sizes included in our analyses should have resulted in more stable estimates of the relationships under investigation. It is also important to point out
that 4 of the 11 articles included in the research by Salas and colleagues were
excluded from our analyses. In one instance, we chose to include a published
version of a dissertation included in the previous integration. In two other
cases, we determined there was no real team-level intervention or team-level
outcomes under investigation. Finally, for a fourth article, only individuallevel outcomes were reported, making that study unusable for our purposes.
These differences in meta-analytic databases, combined with the reality of
using slightly more conservative meta-analysis procedures, may have led to
the divergent findings between the two studies. Specifically, the current
research used meta-analysis methods (i.e., Hunter & Schmidt, 1990; Hunter,
Schmidt, & Jackson, 1982) that can be considered more conservative (Johnson
et al., 1995) than the methods employed by Salas and colleagues (1999),
whose data analyses were based on the methods employed by Rosenthal and
colleagues (i.e., Rosenthal, 1991; Rosenthal & Rubin, 1988).
Implications for Research and Practice
What should academicians and human resources practitioners do when
investigating or recommending team-building interventions? The answers
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
215
seem to lie in the examination of why a team requires a team-building intervention and the characteristics of the team. Human resources practitioners
could play a more proactive role in identifying teams that could benefit
from team building. Specifically, the finding that the role-clarification
and goal-setting components improved performance over the other teambuilding components could benefit human resources practitioners and organizational managers by providing increased clarity into ways in which
leaders may best direct their teams (i.e., being clear about subordinates’
roles and setting goals).
The results of the current meta-analysis provide encouraging news to
human resource practitioners as well as to the many users of team-building
interventions. Generally speaking, the data have suggested that team building has a greater impact on some outcomes over others, and some team
sizes over others. In a case where a manager suspects that his or her team
may benefit from team building it would serve the manager to evaluate and
identify the team’s characteristics, as well as the specific problems encountered prior to intervening with team building. In other words, by communicating to practitioners the varied results that team-building interventions
have on different outcomes, and considering the potential moderating influence of team size, practitioners can be better prepared when assessing what
type of intervention or change effort is most appropriate for their team. The
results from this study reinforce the view that not all teams will benefit
from the same team-building intervention.
Study Limitations
Despite the many interesting findings from the current research integration, there are a number of limitations inherent in this study. First, one conceptual limitation of this study can be summed up by the potentially
interesting and important moderator variables that we did not examine.
These include team type, training duration, criterion source of measurement, and perhaps also a more exacting assessment of the level of interdependence of the teams under investigation in this integration. A proper
analysis of these additional factors may represent a substantial contribution
to the current knowledge base concerning team-building interventions.
There were also a number of methodological limitations inherent to this
research. For one thing, only a fraction of the primary studies reported sufficient data to allow for corrections of the unreliability of criterion
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
216
Small Group Research
measurement. Depending on the analysis in question, only 10%–30% of the
effect sizes in the database had an internal consistency estimate of reliability associated with it. Related to this issue, we took the liberty of assuming
that the predictors, which in this case were the team-building interventions
themselves, were implemented with 100% consistency and accuracy.
Unfortunately, we were not aware of a better estimate of the reliability with
which these interventions are characteristically implemented. Thus, our
assumption and use of perfect predictor reliability is likely an overestimate
of the true reliability of these interventions.
An additional methodological concern in this research concerned the
issue of having a relatively small number of correlations available for many
of the subgroup analyses. Unfortunately, the strict criteria set for inclusion
of primary studies in the database, combined with the relative paucity of
published research in this area, somewhat limited the amount of data
available to us. However, this issue has been relatively common in empirical reviews of team building (e.g., Salas et al., 1999), and the current integration still represents the most exhaustive statement currently available on
this topic.
Directions for Future Research and
What We Still Need to Know
This meta-analysis provides many answers that expand the literature on
team-building interventions. However, there are still some questions that
remain unanswered. Below we present a few recommendations for future
research. Research is needed that examines the efficacy of team building for
various types of teams. Especially needed are published studies of teambuilding interventions for management, and action teams. For example, in
the strategic management literature it has been noted that top management
teams are necessary partners to help set strategic direction, redesign organization architecture, and improve business processes (Allaire, 1998). It would
be well served for future studies to observe the impact that these interventions may have on organizational financial outcomes when implemented in
management teams. Furthermore, future research could also examine if the
reason by which teams are formed has any impact on the efficacy of a teambuilding intervention. If a team is formed because of a directive does team
building affect it differently than a team that was formed by choice?
Additionally, there is a need to further investigate the issue of team size
as it relates to the efficacy of team building. It is an organizational reality
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
217
that organizations require teams of all shapes and sizes. The findings from
this research, where team building appeared to work better for larger teams,
were in direct contrast to those reported by Salas and colleagues (1999).
Concerning the current findings, if team building does affect larger teams
to a greater extent, does it do so only temporarily? Is there a greater need
for follow-up development activities with large teams than small teams?
Could it be that there is an interaction between team size and team type,
such that team size is positively correlated with improvements in team
functioning only for certain team types (e.g., production teams)? For other
teams (e.g., project teams) it may be the case that team building works
better with entities that are smaller in size. This issue could prove to be a
fruitful area of inquiry, but is currently understudied.
Lastly, very few of the studies analyzed here investigated the effects of
team building over time. Instead, most were limited to an isolated postintervention measure of performance. Thus, there is a need for researchers to
investigate the results of team building over the span of the team’s life. It is
in this area that we feel practitioners could make the most impact. It is critical to understand if teams that participate in team building work better as
their time together increases. For example, the longer teams work together
the more likely it is that problems will arise. Future research could assess
if the techniques that teams use to resolve issues are based on what they
learned from their team-building intervention. Ideally, such research would
take pre-intervention measures of performance to calculate a baseline that
would then be followed by multiple measures of performance after the
team-building intervention.
Concluding Remarks
Our findings are encouraging—team building improves team outcomes;
that these team-development interventions are beneficial to team functioning
is the good news. However, we still need to know more about team building.
What are the mechanisms by which it works? What specific features are best,
and how can these be best designed and implemented? In short, more indepth evaluations are needed. If our understanding of the effectiveness and
boundary conditions of team building is to be further enhanced by the type of
evidence-based conclusions derived from meta-analyses—again, we need
more data and more evaluations. This is our call to practitioners of teambuilding interventions to lead the charge in assessing the impact that team
building has on teams in the field. In practice, team building comes in many
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
218
Small Group Research
forms (i.e., games, adventures, or exercises), and is a widely used intervention; this study is an attempt at explaining if it works.
References
References marked with an asterisk (*) indicate studies included in the meta-analysis.
Allaire, P. A. (1998). Lessons in teamwork. In D. C. Hambrick, D. A. Nadler, & M. L.
Tushman (Eds.), Navigating change: How CEOs, top teams, and boards steer transformation (pp. 113-130). Boston: Harvard Business School Press.
Applebaum, E., & Batt, R. (1994). The new American workplace: Transforming work systems
in the United States. Ithaca, NY: IRL Press.
Aronson, E., Ellsworth, P. C., Carlsmith, J. M., & Gonzales, M. H. (1990). Methods of
research in social psychology. Boston: McGraw-Hill.
Bantel, K. A., & Jackson, S. E. (1989). Top management innovations in banking: Does the
composition of the top team make a difference? Strategic Management Journal, 10, 107-124.
Beer, M. (1976). The technology of organization development. In M. D. Dunnette (Ed.),
Handbook of industrial and organizational psychology (pp. 937-994). Chicago: Rand
McNally.
Blake, R. R., & Mouton, I. S. (1964). The managerial grid. Houston: Gulf.
*Boss, R. W., & McConkie, M. L. (1981). The destructive impact of a positive team-building
intervention. Group and Organization Studies, 6, 45-56.
*Bragg, J. E., & Andrews, J. R. (1973). Participative decision-making: An experimental study
in a hospital. Journal of Applied Behavioral Science, 9, 727-735.
Buller, P. F. (1982). The effects of team building-problem solving and goal setting-performance
feedback on the productivity of underground metal miners: A field experiment. Unpublished
doctoral dissertation, University of Washington.
Buller, P. F. (1986). The team building-task performance relation: Some conceptual and
methodological refinements. Group and Organization Studies, 11, 147-168.
*Buller, P. F., & Bell, C. H., Jr. (1986). Effects of team building and goal setting on productivity: A field experiment. Academy of Management Journal, 29, 305-328.
*Bushe, G. R., & Coetzer, G. (1995). Appreciative inquiry as a team-development intervention: A controlled experiment. Journal of Applied Behavioral Science, 31, 13-30.
*Cohen, S. B. (1993). Effects of attribution retraining and positive peer support training on
team cohesion, performance, and participant satisfaction. Unpublished doctoral dissertation, Hofstra University, Hempstead, NY.
DeMeuse, K. P., & Liebowitz, S. J. (1981). An empirical analysis of team-building research.
Group and Organization Studies, 6, 357-378.
*Dionne, S. D. (1998). Team training and development in organizations: A multiple levels of
analysis field experiment. Unpublished doctoral dissertation, State University of New York
at Binghamton, Binghamton.
Dyer, W. G. (1977). Team building: Issues and alternatives. Reading, MA: Addison-Wesley.
Dyer, W. G. (1987). Team building: Issues and alternatives (2nd ed.). Reading, MA:
Addison-Wesley.
*Eden, D. (1986). Team development: Quasi-experimental confirmation among combat
companies. Group and Organization Studies, 11, 133-146.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
219
Field, A. P. (2001). Meta-analysis of correlation coefficients: A Monte Carlo comparison of
fixed- and random-effects methods. Psychological Methods, 6, 161-180.
Field, A. P. (2005). Is the meta-analysis of correlation coefficients accurate when population
correlations vary? Psychological Methods, 10, 444-467.
*Friedlander, F. (1967). The impact of organizational training laboratories upon effectiveness
of ongoing work groups. Personnel Psychology, 20, 289-307.
*Gibson, C. B. (2001). Me and us: Differential relationships among goal-setting training,
efficacy and effectiveness at the individual and team level. Journal of Organizational
Behavior, 22, 789-808.
Gully, S. M., Devine, D. J., & Whitney, D. J. (1995). A meta-analysis of cohesion and performance: Effects of level of analysis and task interdependence. Small Group Research,
26, 497-520.
Halebian, J., & Finkelstein, S. (1993). Top management team size, CEO dominance, and firm
performance: The moderating roles of environmental turbulence and discretion. Academy
of Management Journal, 36, 844-863.
Hambrick, D. C., & D’Aveni, R. A. (1992). Top team deterioration as part of the downward
spiral of large corporate bankruptcies. Management Science, 38, 1445-1466.
*Howard, J. (1979). Organizational team building in church-related teams. Unpublished
dissertation, California School of Professional Psychology, San Diego.
*Huang, W. W., Wei, K., Watson, R. T., & Tan, B. C. Y. (2002). Supporting virtual team-building
with a GSS: An empirical investigation. Decision Support Systems, 34, 359-367.
*Hughes, R. L., Rosenbach, W. E., & Clover, W. H. (1983). Team development in an
intact, ongoing work group: A quasi-field experiment. Group and Organization Studies,
8, 161-186.
Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analysis: Correcting error and bias
in research findings. Newbury Park, CA: Sage.
Hunter, J. E., & Schmidt, F. L. (2004). Methods of meta-analysis: Correcting error and bias
in research (2nd ed.). Thousand Oaks, CA: Sage.
Hunter, J. E., Schmidt, F. L., & Jackson, G. B. (1982). Meta-analysis: Cumulating findings
across research. Beverly Hills, CA: Sage.
Hunter, J. E., & Schmidt, F. L., & Le, H. (2006). Implications of direct and indirect range
restriction for meta-analysis method and findings. Journal of Applied Psychology,
91, 594-612.
Indik, B. P. (1965). Operational size and member participation: Some empirical tests of alternative explanations. Human Relations, 18, 339-350.
Johnson, B. T., Mullen, B., & Salas, E. (1995). Comparison of three major meta-analytic
approaches. Journal of Applied Psychology, 80, 94-106.
*Kimberley, J. R., & Nielsen, W. R. (1975). Organization development and change in organizational performance. Administrative Science Quarterly, 20, 191-206.
Kozlowski, S. W. J., & Ilgen, D. R. (2006). Enhancing the effectiveness of work groups and
teams. Psychological Science in the Public Interest, 7(3), 77-124.
Kraiger, K., Ford, J. K., & Salas, E. (1993). Application of cognitive, skill-based, and affective theories of learning outcomes to new methods of training evaluation. Journal of
Applied Psychology, 78, 311-328.
Light, R. J., & Pillemer, D. B. (1984). Summing up: The science of reviewing research.
Cambridge, MA: Harvard University Press.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
220
Small Group Research
*Longenecker, C. O., Scazzero, J. A., & Stansfield, T. T. (1994). Quality improvement through
team goal setting, feedback, and problem solving. International Journal of Quality and
Reliability Management, 11(4), 45-53.
Macy, B. A., & Izumi, H. (1993). Organizational change, design and work innovation: A
meta-analysis of 131 North American field experiments, 1961-1991. In W. Pasmore &
R. Woodman (Eds.), Research in organizational change and development (pp. 235-313).
Greenwich, CT: JAI.
*Miller, D. J., Jr. (1997). The impact of low ropes course training on team development.
Unpublished doctoral dissertation, Colorado Technical University, Colorado Springs.
*Mitchell, R. (1986). Team building by disclosure of internal frames of reference. Journal of
Applied Behavioral Science, 22, 15-28.
*Morrison, P., & Sturges, J. (1980). Evaluation of organization development in a large state
government organization. Group and Organization Studies, 5, 48-64.
Mullen, B. (1987). Self-attention theory. In B. Mullen & G. R. Goethals (Eds.), Theories of
group behavior (pp. 125-146). New York: Springer-Verlag.
Mullen, B., Anthony, T., Salas, E., & Driskell, J. E. (1994). Group cohesiveness and quality
of decision making: An integration of tests of the groupthink hypothesis. Small Group
Research, 25, 189-204.
Mullen, B., Brown, R. J., & Smith, C. (1992). Ingroup bias as a function of salience, relevance,
and status: An integration. European Journal of Social Psychology, 22, 103-122.
Mullen, B., & Copper, C. (1994). The relation between group cohesiveness and performance:
An integration. Psychological Bulletin, 115, 210-227.
Mullen, B., Salas, E., & Driskell, J. E. (1989). Salience, motivation, and artifact as contributions to the relation between participation rate and leadership. Journal of Experimental
Social Psychology, 25, 545-559.
Noe, R. A. (2002). Employee training and development (2nd ed.). Boston: McGraw-Hill.
Nunnally, J. C. (1978). Psychometric theory. New York: McGraw-Hill.
Payne, V. (2001). The team-building workshop: A trainer’s guide. New York: AMACOM.
Porras, J. I., & Wilkins, A. (1980). Organization development in a large system: An empirical
assessment. Journal of Applied Behavioral Science, 16, 506-534.
Rosenthal, R. (1979). The “file drawer problem” and tolerance for null results. Psychological
Bulletin, 86, 638-641.
Rosenthal, R. (1991). Meta-analytic procedures for social research. Newbury Park, CA: Sage.
Rosenthal, R. (1994). Parametric measures of effect size. In. H. Cooper & L. V. Hedges (Eds.),
The handbook of research synthesis (pp. 231-244). New York: Russell Sage.
Rosenthal, R., & Rubin, D. (1988). Comment: Assumptions and procedures in the file drawer
problem. Statistical Science, 3, 120-125.
Salas, E., & Cannon-Bowers, J. A. (1997). Methods, tools, and strategies for team training.
In M. A. Quinones & A. Ehrenstein (Eds.), Training for a rapidly changing workplace:
Applications of psychological research (pp. 249-280). Washington, DC: American
Psychological Association.
Salas, E., Priest, H. A., & DeRouin, R. E. (2005). Team building. In N. Stanton, H. Hendrick,
S. Konz, K. Parsons, & E. Salas (Eds.), Handbook of human factors and ergonomics methods
(pp. 48-1, 48-5). London: Taylor & Francis.
Salas, E., Rozell, D., Mullen, B., & Driskell, J. E. (1999). The effect of team building on
performance: An integration. Small Group Research, 30, 309-329.
Scargle, J. D. (2000). Publication bias: The “file-drawer” problem in scientific inference.
Journal of Scientific Exploration, 48, 91-106.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
Klein et al. / Team Building
221
Schein, E. H. (1969). Process consultation: Its role in organization development. Reading,
MA: Addison-Wesley.
Schein, E. H. (1999). Process consultation revisited: Building the helping relationship.
Reading, MA: Addison-Wesley.
Schmidt, F. L., & Le, H. (2005). Software for the Hunter-Schmidt meta-analysis methods
(Vol. 1.1). Iowa City: University of Iowa, Department of Management & Organizations.
Shadish, W., Cook, T., & Campbell, D. (2002). Experimental and quasi-experimental designs.
Boston: Houghton Mifflin.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability.
Psychological Bulletin, 86, 420-428.
Stasser, G., Vaughan, S. I., & Stewart, D. D. (2000). Pooling unshared information: The
benefits of knowing how access to information is distributed among group members.
Organizational Behavior and Human Decision Processes, 82, 102-116.
Sundstrom, E., DeMeuse, K. P., & Futrell, D. (1990). Work teams: Applications and effectiveness. American Psychologist, 45, 120-133.
Sundstrom, E., McIntyre, M., Halfhill, T., & Richards, H. (2000). Work groups: From the
Hawthorne studies to work teams of the 1990s and beyond. Group Dynamics, 4, 44-67.
Svyantek, D. J., Goodman, S. A., Benz, L. L., & Gard, J. A. (1999). The relationship between
organizational characteristics and team building success. Journal of Business Psychology,
14, 265-283.
Tannenbaum, S. I., Beard, R. L., & Salas, E. (1992). Team building and its influence on team
effectiveness: An examination of conceptual and empirical developments. In K. Kelley
(Ed.), Issue, theory, and research in industrial/organizational psychology (pp. 117-153).
Amsterdam: Elsevier.
*Wegenast, D. P. (1983). Team building: An application of leadership skills to case management
in child protective services. Unpublished doctoral dissertation, City University of New York,
New York City.
*Wexler, K. J. (1990). Team building: A multidimensional analysis. Unpublished doctoral
dissertation, Hofstra University, Hempstead.
Whitener, E. M. (1990). Confusion of confidence intervals and credibility intervals. Journal of
Applied Psychology, 75, 315-321.
*Woodman, R. W., & Sherwood, J. J. (1980). Effects of team development intervention:
A field experiment. The Journal of Applied Behavioral Science, 16, 211-227.
Cameron Klein works as a survey consultant at Kenexa. Prior to joining Kenexa, Cameron was
employed as an organizational and teams researcher at the Institute for Simulation & Training.
His primary research interests include test development, individual and team training, and interpersonal skills. He has partnered with organizations that include the Army Research Institute,
Kohl’s, Kellogg’s, the National Aeronautics and Space Administration, Ocean Spray, and
Windstream. He is currently a doctoral candidate at the University of Central Florida.
Deborah DiazGranados is a doctoral candidate in the Industrial/Organizational Psychology
program at the University of Central Florida and is a graduate research assistant at the Institute
for Simulation and Training. Ms. DiazGranados received a BS in Psychology and Management
from the University of Houston, and her MS in Industrial/Organizational Psychology from the
University of Central Florida. Her research interests include team processes and effectiveness,
training, motivation, leadership, and the multicultural issues that surround these topics.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009
222
Small Group Research
Eduardo Salas is trustee chair and professor of psychology at the University of Central
Florida where he also holds an appointment as program director for the Human Systems
Integration Research Department at the Institute for Simulation and Training. His expertise
includes assisting organizations in how to foster teamwork, design and implement team-training
strategies, facilitate training effectiveness, manage decision making under stress, and develop
performance measurement tools.
Huy Le is an assistant professor at the Department of Psychology, University of Central
Florida. Dr. Le received his PhD in human resources management at the University of Iowa.
His research interests include personnel selection, test development and validations, crosscultural issues, and quantitative research methodologies (meta-analysis, Monte-Carlo simulations, structural equation modeling).
C. Shawn Burke is a research scientist at the Institute for Simulation and Training at the
University of Central Florida. Her expertise includes teams, team leadership and adaptability,
team training/diagnosis, and team effectiveness. Dr. Burke has over 60 publications related to
the above topics and work accepted at over 100 peer-reviewed conferences. She serves as an
ad hoc reviewer for several journals, including Human Factors, Leadership Quarterly, Journal
of Applied Psychology, and Human Resource Management.
Rebecca Lyons is a doctoral student in the Industrial and Organizational Psychology program
at the University of Central Florida. She earned a BS in Psychology in 2004 from Davidson
College. Rebecca is a graduate research associate at the Institute for Simulation and Training
where her research interests include individual and team training, performance measurement,
multiteam systems, and simulation, with an emphasis in healthcare.
Gerald F. Goodwin is a research psychologist at the U.S. Army Research Institute (ARI) for
Behavioral and Social Sciences. He received his MS and PhD in Industrial/Organizational
Psychology from Pennsylvania State University. Dr. Goodwin is currently a special projects
officer in ARI’s Strategic Initiatives Group, with responsibility for topics related to ARI’s
leader and organization research programs. Dr. Goodwin’s research expertise is in leadership,
team effectiveness, and organizational issues in joint, interagency, and multinational contexts.
Downloaded from http://sgr.sagepub.com at Goteborgs Univ.- bibliotek on May 18, 2009